Use Physionet EHG preterm/term database to make a classifier. The classifier is able to classify whether the patient will have a preterm or a term labor, using EHG signal and past medical history features, with an average sensitivity of 92.6%.
More in-depth project description, including our research and design approach.
Physionet database: https://physionet.org/physiobank/database/tpehgdb/
We are using Python 3x, pip, and pipenv.
Pipenv installation:
pip install pipenv
Once you have Python 3x, pipenv, and the repo downloaded, go to the root directory and initialize the pipenv and all of the dependencies:
pipenv install
Running different files:
pipenv run .\[filename]
Running the main.py will print the accuracy, sensitivity, and specificity of our best classifier, model 6, and output a csv file with the sensitivity and specificity for each of the 30 trials.
pipenv run .\main.py
Implement Random Forest, logistic regression, and other ML models
- pipenv - The virtual environment
- wfdb - For downloading the Physionet's dataset
- tensorflow - For building and evaluating the neural network
- imbalanced-learn - Used to balance the dataset
- numpy - Used for fomatting the data
- scikit-learn - Used for splitting the data
- Dian Guo
- Michael Isaf
- Alexandra Melehan
- Brandi Nevius